19:03:54 <jeblair> #startmeeting ci
19:03:55 <openstack> Meeting started Tue Dec 18 19:03:54 2012 UTC.  The chair is jeblair. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:03:56 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:03:58 <openstack> The meeting name has been set to 'ci'
19:04:18 <jeblair> #topic wiki
19:04:32 <jeblair> Ryan_Lane: ping
19:04:38 <Ryan_Lane> pong
19:04:49 <jeblair> so we were just talking about this in openstack-infra
19:05:16 <jeblair> seems like there are a few things that need sprucing up before we make the switch
19:05:30 <Ryan_Lane> the biggest thing I can see is the lack of a theme
19:05:53 <Ryan_Lane> and rewrite rules
19:05:56 <fungi> it's always all about aesthetics with some people ;)
19:06:14 <Ryan_Lane> I don't think we should make any content changes until the wikis switch
19:06:28 <jeblair> Ryan_Lane: what do you mean?
19:06:41 <Ryan_Lane> do a straight migration
19:06:48 <Ryan_Lane> then when we switch wikis, do the style fixes and such
19:07:04 <Ryan_Lane> otherwise we'll have edits occurring in both at the same time
19:07:17 <Ryan_Lane> and we'll end up losing edits
19:07:34 <jeblair> Ryan_Lane: that makes sense, but there are some things we can take care of before hand, like the theme, right?
19:07:35 <Ryan_Lane> wiki transitions aren't easy and often take months for everything to be perfect
19:07:38 <Ryan_Lane> yes
19:07:39 <clarkb> can we make the old wiki read only?
19:07:45 <Ryan_Lane> yep
19:07:58 <clarkb> would that help avoid losing edits?
19:08:21 <jeblair> clarkb: i don't think they have to run in parallel for very long or at all
19:08:22 <Ryan_Lane> yep, but that would mean we've made the choice to switch at the point
19:08:35 <clarkb> I see
19:08:46 <jeblair> clarkb: when we are ready for the switch, we can take a short outage to switch the hostnames and then we're done...
19:08:54 <jeblair> Ryan_Lane: how long did the import/fixup take?
19:09:01 <Ryan_Lane> about an hour
19:09:10 <Ryan_Lane> well, maybe 2
19:09:34 <jeblair> that might be long enough to leave the old wiki running read only while that's going on
19:09:42 <jeblair> and then do the hostname switch near the end of that time
19:09:55 <jeblair> though i don't think a little bit of downtime is going to kill us.
19:10:22 <jeblair> but before we make the switch, i think we should have the theme in place, and probably solutions to the major problems...
19:10:45 <jeblair> ie, we should know how we're going to fix big things like the main page
19:10:49 <Ryan_Lane> yeah. we'll need to find someone to do the theme.
19:11:25 <jeblair> Ryan_Lane: what's involved in a mediawiki theme?
19:11:32 <Ryan_Lane> writing a skin
19:11:43 <Ryan_Lane> we may have to pay someone for it.
19:11:50 <clarkb> I would volunteer but I have zero skill in making things look pretty
19:11:57 <Ryan_Lane> that's my problem, too ;)
19:12:12 <Ryan_Lane> I'm very much not a front-end dev
19:12:45 <zykes-> clear
19:13:04 <jeblair> Ryan_Lane: or ask for volunteers on the list?
19:13:16 <clarkb> jeblair: looks like he just did :)
19:13:30 <jeblair> ok!
19:14:08 <Ryan_Lane> I can ask the folks who made the webplatform.org skin
19:14:18 <Ryan_Lane> they seemed to do a good job, but they were paid contractors
19:15:08 <jeblair> ah.  our current skin doesn't look like much more than a logo and some colors.  it probably doesn't have to be too fancy.  :)
19:16:11 <Ryan_Lane> yep
19:16:16 <jeblair> Ryan_Lane: what about image inclusion?
19:16:38 <Ryan_Lane> the migration script doesn't handle uploads properly
19:16:47 <Ryan_Lane> I can pull all the images and upload them via a script
19:16:56 <Ryan_Lane> but we'll need to fix all the places they are included
19:17:34 <Ryan_Lane> other syntax fixes shouldn't take ages
19:17:50 <Ryan_Lane> last time I tried this, it took me 3 hours to clean up almost all of the linked articles
19:18:00 <Ryan_Lane> a sprint after we switch over should fix most things pretty quickly
19:18:08 <Ryan_Lane> we'll need a youtube extension
19:18:15 <Ryan_Lane> since mediawiki doesn't really have one
19:18:25 <Ryan_Lane> I may be able to write one fairly quickly
19:18:43 <jeblair> annegentle_: ^
19:19:06 <jeblair> maybe annegentle_ could help with the sprint
19:19:35 <Ryan_Lane> yeah
19:20:16 <jeblair> #topic things to do before final wiki migration
19:20:27 <jeblair> #info openstack theme skin
19:20:41 <jeblair> #info image upload script
19:20:47 <jeblair> #info youtube extension
19:21:06 <clarkb> #info organize content cleanup sprint
19:21:35 <jeblair> #info figure out how to make the front page look pretty
19:21:48 <jeblair> #info rewrite rules for content at /
19:21:51 <jeblair> anything else?
19:22:09 <Ryan_Lane> I think that's about it
19:22:26 <jeblair> cool, thanks!
19:22:36 <jeblair> I have the new cert in hand, i'll try to get it into puppet today
19:23:12 <jeblair> #topic cla project
19:23:31 <fungi> updates...
19:23:36 <fungi> #link https://etherpad.openstack.org/cla-maintenance-2012
19:23:41 <jeblair> jbryce is supposed to send me the text of the new cla soon
19:23:46 <fungi> i've got an implementation sequence there
19:23:58 <fungi> and a few items still to do before we can schedule
19:24:06 <fungi> the cla text is one of them
19:24:34 <fungi> since we've muddied this with getting rid of lp group sync, i have yet another question related to that
19:24:58 <fungi> specifically, what's going to be the duplication in group management at that point?
19:25:20 <fungi> i gather there will still need to be some groups on lp for things which lp needs to use in its own acls
19:25:39 <jeblair> fungi: basically the -drivers teams has meaning in both places, so likely has duplicate group management.
19:25:41 <clarkb> I think the only stuff on lp would be things tied to blueprints and bugs
19:25:49 <fungi> and do we want those handled separately, or do we try to engineer some wort of reverse sync?
19:26:15 <fungi> clarkb: yeah, those were the only two which came to mind, so they're in the etherpad there
19:26:23 <jeblair> -drivers changes very slowly, so i think we don't sync, just duplicate manual work.
19:26:43 <fungi> i'm good with that as long as everyone else is
19:27:03 <jeblair> -drivers gets to set milestone targets.  we don't do anything with the bugs group in gerrit (in lp it can assign bugs).  and we don't do anything with -core in lp
19:27:19 <fungi> might encourage reddwarf to scale back on their 40+ drivers
19:27:29 <jeblair> yes, they are doing it wrong.
19:27:32 <fungi> if they had to maintain that list by hand in two places
19:28:11 <jeblair> so close!  anything else on this?
19:28:13 <jeblair> mordred: ping
19:28:19 <fungi> okay, so -drivers lives in both lp and gerrit, -core is only gerrit and -bugs is only lp
19:28:51 <fungi> i've added an additional e-mail text there, which is the reminder to go out after the maintenance comcludes
19:28:54 <fungi> er, concludes
19:29:09 <fungi> because there *will* be devs taken by surprise
19:29:25 <fungi> many don't read -dev@ either, so it's inevitable anyway
19:29:28 <clarkb> do we have a targetted date for the switch? I assume not before the new year?
19:29:55 <fungi> at this point we're so close to the end of the year it seems likely i'll rename that etherpad to -2013 :(
19:30:21 <jeblair> heh.  let's get the cla text and everything in hand before we set a date
19:30:51 <fungi> that's my feeling as well. don't want to go shifting dates on announcements and adding to the already expected confusion
19:31:08 <fungi> i want all the ducks in a row before i send the announcement
19:31:13 <clarkb> ++
19:31:49 <fungi> anyway, that's where we stand
19:32:16 <fungi> patches have been rebased, keys added in hiera, et cetera
19:32:27 <annegentle_> jeblair: Ryan_Lane hey I was going to ask if we should just do a wiki format day
19:32:39 <Ryan_Lane> yeah, we should
19:32:45 <fungi> some remaining doc updates i need to finish muddling through (meant to be working on those last night and this morning)
19:32:58 <Ryan_Lane> it only took me a few hours to fix most of the glaring issues last time I tried it
19:33:08 <jeblair> annegentle_: yeah, and probably the closer to the transition the better.  like maybe the day of? :)
19:33:59 <jeblair> annegentle_: but we probably shouldn't schedule that until we have the other issues we identified worked out.
19:34:22 <jeblair> fungi: sounds good, thanks!
19:34:30 <fungi> my pleasure
19:34:32 <jeblair> #topic askbot
19:34:48 <jeblair> reed brought my attention to this: https://twitter.com/morganmclean/status/280833638864654336
19:35:35 <jeblair> and suggested that increasing the prioritization of the provisioning of an askbot server might be nice.  :)
19:35:53 <jeblair> mordred was going to take a first stab at this, but i don't think that has happened.
19:36:29 <jeblair> i could try to do it later this week
19:36:39 <reed> that would be good
19:36:55 <clarkb> jeblair: are you planning to stand the whole thing up or just provision the host?
19:37:11 <clarkb> I seem to recall there was a dev type install somewhere
19:37:11 <jeblair> if anyone else would like to volunteer, that would be swell.
19:37:44 <clarkb> what is the scope of the work that needs to be done?
19:39:13 <jeblair> that is not clear to me.  i think mordred's plan was to set up a host that installed askbot from a git repo somewhere?
19:39:20 <jeblair> i don't recall all the details.
19:40:08 <clarkb> ok, so some scoping would need to be done. I don't want to volunteer now as I really need to get on top of reviews and want to be open enough to fight testr fires should they occur
19:40:33 <clarkb> but maybe late thursday or early friday we can do a status check and I may be able to start that
19:41:13 <jeblair> reed: do you know which askbot wants to be installed?
19:42:15 <jeblair> http://askbot.org/doc/install.html
19:43:00 <reed> jeblair, that one
19:43:44 <reed> I can put you in touch with the people that will customize it
19:46:42 <clarkb> I was wondering if something like that happened
19:46:47 <clarkb> must've lost power.
19:46:48 <fungi> i guess the weather finally got him
19:46:59 <clarkb> anything else to go over about askbot?
19:46:59 <ttx> did we get to the bottom of the tarball job failures yet ?
19:47:10 <fungi> ttx: yes, we did
19:47:16 <clarkb> #topic tarball job failures
19:47:28 <ttx> fungi: nova-tarball was still failing 52 minutes ago
19:47:32 <fungi> #link https://etherpad.openstack.org/ci-issues-log
19:47:39 <clarkb> need to be jeblair or meetbot super user :(
19:47:42 <fungi> #link https://review.openstack.org/18334
19:48:16 <fungi> clarkb: you want to review 18334 so ttx can haz tarballs?
19:48:25 <ttx> I hope jeblair will come back in time to #endmeeting :)
19:48:30 <clarkb> yes, I will move that to the front of the list
19:48:38 <ttx> fungi: not so much of a hurry
19:48:42 <clarkb> ttx: if he doesn't I can get meetbot suoper user I think
19:48:43 <ttx> It's ok if it's under control
19:49:05 <fungi> ttx: it was an unintended outcome of the zuul move over the weekend
19:49:06 <ttx> no need to super-prioritize it, I don't need tarballs just now
19:49:10 <clarkb> k
19:49:26 <fungi> fairly quickly rooted out once jeblair got online
19:49:44 <fungi> i was looking in the right place, but the fix was a bit beyond me
19:49:55 <clarkb> what else have we got? nova testr happened
19:50:06 <fungi> boy howdy did it
19:50:26 <clarkb> overall seems to be going relatively smoothly. jog0 discovered testr didn't run on OS X, lifeless quickly sorted that out
19:50:44 <clarkb> there were also a few, previously undiscovered, inter test conflicts
19:50:53 <clarkb> lifeless and I have fixed two of those
19:51:02 <fungi> one nova dev found a nice way to persistently dos our jenkins slaves with it
19:51:19 <fungi> though i think similar forkbombs would have been possible before
19:51:24 <clarkb> fungi: I am not completely sure the same wouldn't happen with nosetests though
19:51:28 <clarkb> ya
19:51:29 <fungi> right
19:51:45 <fungi> however, we upgraded their pistols to rocket launchers
19:52:40 <fungi> still not certain what the best resource limit fix is there
19:52:46 <clarkb> there has been some feedback about whether or not defaulting to --parallel is what everyone wants. It will definitely need to grow and mature as we use it. I think that jumping into the deep end was the only real sane way to sort these things out though
19:53:24 <clarkb> in addition to the OS X + testr problem lifeless discovered that python discover does not return test lists in a deterministic order
19:53:31 <fungi> no argument there. but i believe (for example) sdague pointed out that having tests launch multiple vms in parallel will be resource-swamping
19:53:48 <clarkb> this makes local test reproduction difficult, so he has it on his plate to work around that and has submitted a bug to upstream discover
19:54:08 <lifeless> I've also put a workaround in testtools
19:54:35 <clarkb> lifeless: has that been released to pypi? eg will rebuilding venvs at this point result in deterministic test reproduction?
19:55:04 <clarkb> (its ok if it hasn't just trying to keep on top of it)
19:55:10 <lifeless> no :)
19:55:25 <clarkb> thats all I had for nova testr
19:55:34 <lifeless> will be either late today or first thing tomorrow, I may have other testtools things to land first
19:55:47 <clarkb> I am going to start trying to escalate my meetbot permissions
19:55:47 <lifeless> and its not a panic-thing, though good to have.
19:56:23 <clarkb> fungi: ttx mordred any other things we want to go over?
19:56:28 <ttx> nope
19:56:31 <fungi> i guess we're at open discussion now in jeblair's absence. nothing else from me
19:56:46 <jeblair> #endmeeting